|
Network with the computer more and more widely used in the work of today 's social, business, government and other organizations are increasingly dependent on computer network systems , security issues become more prominent . Intrusion detection system IDS (Intrusion Detecton System) has become an important means essential . In order to overcome the shortcomings of existing ID models or products , from ID standardization apply neural network NN (Neural Network) ID . How to detect the increasing spread of the invasion , has been beyond the capacity of any one IDS products . Collected raw data , how to effectively analyze and report the results of these data has been the focus of research in the field , and therefore the formation of a variety of ID . Based on the above research , the paper proposes an ID model , that is, a neural network - based intrusion detection model . Introduces a neural network model , ID standardization research . We studied the realization of the model , and proposes an improved training algorithm , Finally , a core component of the model analysis , design and implementation , and related experiments . Experiments show that application of neural networks to intrusion detection , can achieve better training results , with improved the BP algorithm detection accuracy of up to 91.5% , the lowest false negative rate to 3.5% , the lowest rate of false positives to 4.5% , for the real-time is not a high system requirements , it is better able to meet the need .
|